Introduction

The race to develop AI and calls to regulate it are heating up around the world. In November 2023, 28 countries, including the United States, United Kingdom, China, and the European Union, met and pledged to work together to address and contain the potentially harmful risks of AI. These countries are establishing guidelines to manage the risks, while balancing the benefits of this new and evolving technology.

AI holds the promise of dramatic improvements across industries and greater cybersecurity by improving threat detection and identifying anomalies. Yet, AI also brings potential risks to personal privacy, job loss, surveillance, bioweapons, cyberattacks, intellectual property issues and the proliferation of misinformation, among other challenges.

Global Risks for Business

1. Risks and costs associated with compliance across multiple jurisdictions
   Among the legislation being introduced in different jurisdictions, the EU is in the final stages of formulating its AI Act. The legislation would establish requirements for providers and users depending on the level of risk, whether it be “unacceptable risk,” which means those systems considered a threat to people will be banned; or “high risk,” which would adversely affect fundamental rights or safety. The EU is aiming for a more comprehensive approach to AI regulation in an attempt to limit surveillance while the US is seeing more of a patchwork and has yet to create a unified regulatory approach throughout its various states. Still, US President Joe Biden issued an executive order in late 2023 focusing on developing AI in tandem with safety and security. According to the White House, it establishes new standards for protecting the privacy and interests of both consumers and workers, and advances equity and civil rights while promoting innovation and competition.
2. AI for criminal activity
   Generative AI (Gen-AI), like ChatGPT and other chatbots, can imitate human speech and writing to create phony personas and documents, as well as “deep fake” photos and videos that have shown prominent people in false and reputationally harmful settings. Bot networks can create new ways to exploit vulnerabilities, including the expansion of phishing to gain access to personal information, as well as the replication of voices to fraudulently request money from family and friends. These uses of AI tools by threat actors to facilitate wrongdoing pose a growing risk to both businesses and consumers.
3. AI and intellectual property issues
   When Gen-AI creates a particular work, there is a question as to whether copyright, patent, or trademark protections apply. There is also uncertainty pertaining to the ownership of AI-generated works. Additionally, tech companies have increasingly been sued for using copyrighted material to train their AI programs. Some recent examples include a class action filed by a San Francisco-based law firm challenging the training and output of AI-based systems for software programmers. More recently, the New York Times filed a lawsuit over the use of millions of its articles to train ChatGPT without the newspaper’s permission.
4. ChatGPT and AI hallucinations
   AI is not infallible, especially when it is not trained with a sufficiently robust dataset that is tested for bias. As a result, AI can produce hallucinations, namely answers that are not accurate or are completely false. AI hallucinations pose the risk of reputational damage or criminal charges. In one example, a US lawyer was sanctioned by a court for using ChatGPT for legal research which was full of false case citations. As a result, at least two federal district courts are mandating certification of the use of AI in cases, and a federal appeals court is considering doing the same. Under certification requirements that have been adopted or proposed, lawyers must attest that either no portion of the filing was drafted using Gen-AI, or if Gen-AI was relied upon, that the work was checked for accuracy by a human being.
5. Increased dependency on AI

Automation, especially with the increasing adoption of Gen-AI, will create dependency, which combined with the loss of skills could result in a multiplying effect on the ability to deliver proactive and responsive measures. Once dependency has been “baked in” to a workflow, it may be very difficult to correct. For example, poorly programmed AI could result in biased outputs. Gen-AI could also be used as a crutch that hinders critical thinking and creativity of real people.

Global Opportunities for Business

1. Fraud detection
   AI can help legal and regulatory consultants to detect fraud and conduct investigative data analytics. Compliance departments and legal teams can use Gen-AI systems to examine financial transactions in real-time to uncover fraud and initiate internal investigations or examine practices of external vendors or sales agents. Regulators are encouraging many companies to expand internal compliance review capabilities with advanced analytics systems, inclusive of machine learning AI, which can predict patterns of wrongdoing.
2. Enhanced data management
   Gen-AI can be used to provide enhanced data management practices and information governance by improving classification according to different types of media; increasing the quality of data by reducing errors; keeping data secure from threat actors, while in compliance with laws and regulations; locating data; and integrating information from various lists and sources.
3. AI can bolster cybersecurity
   AI can be integrated into cybersecurity workflows to enable the quick triaging of alerts, automated incident response workflows, predictive review for phishing detection, identifying bots, and flagging suspicious access.
4. AI and litigation document review
   Language Processing AI, such as ChatGPT, is starting to be introduced to discovery processes of litigation. The eDiscovery industry often uses Technology Assisted Review and other analytic methods to find the most relevant documents early in a review, thereby reducing the cost of the process.
5. Insurance industry benefits

From underwriting to customer service and telematics, the insurance industry has been a leader in the innovative uses of AI. AI can input claims data faster over a long period of time by enabling the automation of time-consuming and repetitive tasks, such as data entry and document verification. It can take photos of car accident scenes and use telematics data to determine how the actions of the drivers or even the road conditions contributed to the incident. AI is accelerating fraud investigations by its ability to analyze vast amounts of data in real-time and quickly identify suspicious patterns. AI is helping the insurance industry to forecast natural disasters and extreme weather events in addition to quickly processing claims.

Top Takeaways

1. 1. Legislation focused on developing AI while mitigating its potential risks is being planned worldwide. While there is no single global regulatory authority that will harmonize the various AI laws, there is a growing patchwork of AI rules and regulations.
   2. Developers will be required under new laws to be more transparent about the risks that AI presents and, in some cases, will have to submit their technology to government agencies for risk assessments.
   3. Potentially conflicting laws between the EU, UK, US, China, and other countries means that maintaining cross-border compliance might prove costly and could outweigh the benefits of its use.
   4. Companies will face difficulties harmonizing AI regulations with environmental, social, and governance (ESG) laws and other rules, especially in the area of supply chain compliance.
   5. Tech businesses will play an increasingly important role alongside governments in both developing new AI applications and identifying and limiting potential risks.